similar to: Override TargetOptions for block of code?

Hello everyone, I'd like to propose the attached patch to form FMA intrinsics aggressively, but in order to do so I need some clarification on the intended semantics for the various FP precision-related TargetOptions. I've summarized the three relevant ones below: UnsafeFPMath - Defaults to off, enables "less precise" results than permitted by IEEE754. Comments specifically

Hi Owen, Having looked into this due to Clang failing PlumHall with it recently I can give an opinion... I think !NoExcessFPPrecision covers FMA completely. There are indeed some algorithms which give incorrect results when FMA is enabled, examples being those that do floating point comparisons such as: a * b + c - d. If c == d, it is still possible for that result not to equal a*b, as "+c

To continue the discussion I started last year (see the link below) on embedding command-line options in bitcode, I came up with a plan to improve the way the backend changes UnsafeFPMath on a per-function basis. The code in trunk currently resets TargetOptions::UnsafeFPMath at the beginning of SelectionDAGISel::runOnMachineFunction to enable compiling one function with “unsafe-fp-math=true” and

Hi, Our DSL LLVM IR emitted code (optimized with -O3 kind of IR ==> IR passes) runs slower when executed with the LLVM 3.3 JIT, compared to what we had with LLVM 3.1. What could be the reason? I tried to play with TargetOptions without any success… Here is the kind of code we use to allocate the JIT: EngineBuilder builder(fResult->fModule);

On Thu, Jul 18, 2013 at 9:07 AM, Stéphane Letz <letz at grame.fr> wrote: > Hi, > > Our DSL LLVM IR emitted code (optimized with -O3 kind of IR ==> IR passes) runs slower when executed with the LLVM 3.3 JIT, compared to what we had with LLVM 3.1. What could be the reason? > > I tried to play with TargetOptions without any success… > > Here is the kind of code we use to

----- Original Message ----- > From: "Reid Kleckner" <rnk at google.com> > To: "Akira Hatanaka" <ahatanak at gmail.com> > Cc: "LLVM Developers Mailing List" <llvmdev at cs.uiuc.edu> > Sent: Monday, January 12, 2015 2:54:48 PM > Subject: Re: [LLVMdev] Enable changing UnsafeFPMath on a per-function basis > > > > Whatever

Le 18 juil. 2013 à 19:07, Eli Friedman <eli.friedman at gmail.com> a écrit : > On Thu, Jul 18, 2013 at 9:07 AM, Stéphane Letz <letz at grame.fr> wrote: >> Hi, >> >> Our DSL LLVM IR emitted code (optimized with -O3 kind of IR ==> IR passes) runs slower when executed with the LLVM 3.3 JIT, compared to what we had with LLVM 3.1. What could be the reason? >>

On Wed, 2012-02-08 at 10:11 -0800, Owen Anderson wrote: > Hello everyone, > > > I'd like to propose the attached patch to form FMA intrinsics > aggressively, but in order to do so I need some clarification on the > intended semantics for the various FP precision-related > TargetOptions. I've summarized the three relevant ones below: > > > UnsafeFPMath -

I realize that some of these thoughts apply equally to the prior !fpaccuracy metadata that's been around a while, but I hadn't looked at it closely until now. The !fpmath metadata violates the original spirit of metadata, which is that metadata is only supposed to exclude possible runtime conditions, rather than to introduce new possible runtime possibilities. The motivation for this is

Hi Dan, > I realize that some of these thoughts apply equally to the > prior !fpaccuracy metadata that's been around a while, but I > hadn't looked at it closely until now. > > The !fpmath metadata violates the original spirit of > metadata, which is that metadata is only supposed to exclude > possible runtime conditions, rather than to introduce new > possible

Hi Duncan, I'm not an expert in fp accuracy question, but I had quite a few experience dealing with fp accuracy problems during compiler transformations. I think you have a step in the right direction, walking away from ULPs, which are pretty useless for the purpose of describing allowed fp optimizations IMHO. But using just "fast" keyword (or whatever else will be added in the

Hi, We 've been working on an LLVM backend for High Performance Erlang (HiPE) [1], the native code compiler of Erlang/OTP [2]. ErLLVM [3] targets the X86 and AMD64 architectures for now but there is some ongoing work from a team on the Uppsala University to also support ARM. In our implementation, we have paid special attention on retaining ABI-compatibility with the Erlang Runtime System in

Hi Dmitry, > I'm not an expert in fp accuracy question, but I had quite a > few experience dealing with fp accuracy problems during compiler transformations. I agree that it's a minefield which is why I intend to proceed conservatively. > I think you have a step in the right direction, walking away from ULPs, which > are pretty useless for the purpose of describing allowed

Hi all, we have several features in Mips that are dependent on target abi. A recent commit(r224492) introduced a new -target-abi option to TargetOptions struct that provides access to abi string. This info is stored in MCSubtarget class(ARMSubratget in this case) and distributed to other libraries. Unfortunately, for Mips we need this info in other classes that don't have access to MCSubtarget

Duncan, I have some issues with representing this as a single "fast" mode flag, which mostly boil down to the fact that this is a very C-centric view of the world. And, since C compilers are not generally known for their awesomeness on issues of numerics, I'm not sure that's a good idea. Having something called a "fast" or "relaxed" mode implies that it is

The attached patch is a first attempt at representing "-ffast-math" at the IR level, in fact on individual floating point instructions (fadd, fsub etc). It is done using metadata. We already have a "fpmath" metadata type which can be used to signal that reduced precision is OK for a floating point operation, eg %z = fmul float %x, %y, !fpmath !0 ... !0 = metadata

Hi Eric, The main thing we need to fix is that the selection between ELF32/ELF64 needs to depend on the ABI being N64 and not on whether we used a mips-linux-gnu triple versus a mips64-linux-gnu triple. So 'clang -target mips-linux-gnu' -mips64r2 -mabi=64' should produce an ELF64 and 'clang -target mips64-linux-gnu -mips32r2 -mabi=32' should produce an ELF32. In terms of code,

Hi Eric, as I was working on the same issues that are covered in your patch I also made a change in clang driver to pass this option to the assembler. Could you please review it and tell me your opinion? http://reviews.llvm.org/D6091 Thanks Vladimir ________________________________ From: Daniel Sanders Sent: Wednesday, January 28, 2015 8:59 PM To: Eric Christopher; Vladimir Medic; llvmdev at

On Jun 13, 2008, at 12:27 AM, Nicolas Capens wrote: > Hi all, > > When trying to generate a VFCmp instruction when UnsafeFPMath is set > to true I get an assert “Unexpected CondCode” on my x86 system. This > also happens with UnsafeFPMath set to false and using an unordered > compare. Could someone look into this? Have you filed a bug? > > While I’m at it, is there

Hi Chris, I've attached a replacement of fibonacci.cpp that reproduces the issue on x86/SSE systems. Regarding the definition of the VFCmp instruction, I think it would really be a lot more valuable to define it as returning all 1's or all 0's per element. Just setting the most significant bit is pretty much worthless (someone correct me if I'm wrong). I checked and I